Faculty Bibliography

Introduction: Sotorasib is a first-in-class small molecule that specifically and irreversibly inhibits KRAS^G12C. In the registrational phase 2 CodeBreaK 100 trial, sotorasib showed an objective response rate (ORR) of 37.1% and a median progression-free survival (PFS) of 6.8 months in patients with KRAS p.G12C-mutated non-small cell lung cancer (NSCLC) previously treated with platinum-based chemotherapy and/or immunotherapy. Response to sotorasib has been observed across co-occurring mutational profiles. Here, we report preliminary data on the genomic profiles and potential determinants of response to sotorasib from an exploratory analysis of this trial.
Method(s): Baseline tissue samples were collected and analyzed for genomic alterations in KEAP1, the upstream RTK pathway, and the downstream PI3K/AKT/mTOR and MAPK pathways. Patients were categorized into the following 3 groups: early progressors (patients with an event of progressive disease and PFS of <3 months), late progressors (patients with an event of progressive disease and PFS of >= 3 months), non-progressors (patients with no event of progressive disease and PFS of >= 3 months).
Result(s): A total of 126 patients were enrolled into the phase 2 trial. 65 patients with available data from baseline tissue samples were categorized per methods: 22 early progressors, 23 late progressors, and 20 non-progressors. 11 of the 65 patients had KEAP1 mutations (7 in the early progressor group, 2 in the late progressor group, and 2 in the non-progressor group). In the early progressor group, we observed mutations in EGFR, FGFR, PDGFR, RET, and MET, which were also identified in other groups. Among the patients with mutations in the MAPK pathway genes, the late progressor group was the most prevalent (43%, n=6), followed by early progressor (29%, n=4) and non-progressor (29%, n=4). Among patients with mutations in genes of the PI3K/AKT/mTOR pathway, late progressor and early progressor groups were the most prevalent (36%, n=9, each), followed by non-progressor group (28%, n=7) (summary Table below). [Formula presented]
Conclusion(s): In this descriptive biomarker analysis of baseline tissue specimens from the phase 2 CodeBreaK 100 trial of sotorasib in KRAS p.G12C-mutated NSCLC, diverse mutation patterns were observed. No unique genomic profiles were identified in patient groups. The presence of KEAP1 mutation was observed across all groups and was more prevalent in early progressors. These findings warrant further investigation of the longitudinal cfDNA dynamics in patients receiving sotorasib. Keywords: biomarkers, sotorasib (AMG 510), KRAS p.G12C NSCLC
Copyright

Introduction: Gemcitabine and carboplatin/cisplatin ("platinum")-based combinations are used to treat a wide variety of malignancies including gynecologic, breast, lung, and occult primary cancers. In Non-Small Cell Lung Cancer (NSCLC), these combinations led to a substantial improvement in overall survival. Nevertheless, a large proportion of patients do not respond. An optimal cytotoxic strategy for managing NSCLC and the discovery of predictive biomarkers for cytotoxic chemotherapy to guide treatment selection remain unmet needs in the clinic. The Cellworks Computational Omics Biology Model (CBM) platform identified a unique chromosomal signature which permits a stratification of patients that are most likely to respond to gemcitabine and platinum treatments.
Method(s): Twenty patients treated with gemcitabine and platinum were identified from a TCGA dataset and analyzed. The mutation and copy number aberrations from individual cases served as input into the CBM to generate a patient-specific protein network map from PubMed and other online resources. Disease-biomarkers unique to each patient were identified within patient-specific protein network maps. Digital drug biosimulations were conducted by measuring the effect of gemcitabine and platinum on a cell growth score comprised of a composite of cell proliferation, viability, apoptosis, metastasis, and other cancer hallmarks. Drug biosimulations were conducted by mapping the drug combination to the patient genome along with a rational mechanism of action and validated based on the patient's genomic profile and biological consequences.
Result(s): Of the 20 patients treated with gemcitabine and platinum, 12 had clinical responses while 8 were non-responders. The CBM correctly predicted response in 17/20 patients with 85% accuracy, 63% specificity and 100% sensitivity. The CBM identified that novel amplified segments of Chromosome 6p were associated with non-responsiveness to gemcitabine and platinum therapy. Key genes on these segments include E2F3, MDC1, TAP1 and TNF. Amplification of E2F3 leads to activation of MSH2/6, which enhances mismatch repair thereby causing resistance. Amplification of MDC1 leads to activation of CHECK2, BRCA1, ATM, and NBN_RAD51_MRE1 Complex which stimulates homologous recombination repair. Amplification of TAP1 reduces gemcitabine transport. Besides 6p amplification, PRMT7 deletion was also associated with gemcitabine resistance. Notably, BRCA2-del, RB1-del, NPM1-del, LIG4-del, XRCC4-del, RAD50-del, ATRX-Del, RBBP8-del, XRCC6-del, and FBXW7-del were also prevalent among gemcitabine non-responders. Interestingly, these aberrations also happen to be key criteria for predicting response to etoposide. Therefore, etoposide and platinum combinations might have provided better disease control for these patients.
Conclusion(s): Amplification of chromosome 6p appears to be an important cause of treatment failure for patients receiving gemcitabine-platinum combinations. In this small patient group, the Cellworks CBM was especially useful for identifying non-responders. Biosimulation can identify novel patient subgroups for therapy response prediction and has promise to help select more effective therapies. Keywords: Multi-omics Therapy Biosimulation, Personalized Cancer Therapy, Cancer Therapy Biosimulation
Copyright

Introduction: Cytotoxic drugs are hampered by limited efficacy. Hence, a personalized treatment approach matching chemotherapy with appropriate patients remains an unmet need. Genomic heterogeneity creates an opportunity to discern key genomic aberrations and pathways that confer resistance and response to standard treatment options. We conducted a study using the Cellworks Computational Omics Biology Model (CBM) to identify novel genomic biomarkers associated with response among Non-Small Cell Lung Cancer (NSCLC) patients receiving platinum-based treatments.
Method(s): 104 NSCLC patients who received platinum-based chemotherapy were selected from TCGA: platinum-etoposide (N=18), platinum-gemcitabine (N=20), platinum-vinorelbine (N=31), platinum-paclitaxel (N=21), and platinum-docetaxel (N=14). Mutation and CNV from each case served as input for the CBM to generate a patient-specific protein network-map based on PubMed and other resources. Biomarkers unique to each patient were identified within protein network-maps. Drug impact on the disease network was biosimulated to determine efficacy score by measuring the effect of chemotherapy on the cell growth score, a composite of cell proliferation, viability, apoptosis, metastasis, DNA damage and other cancer hallmarks. Effectively, the mechanism of action of each drug was mapped to each patient's genome and biological consequences determined response.
Result(s): Among the 104 patients, 74 were responders (R) and 30 non-responders (NR), determined using compete and partial response based on RECIST criteria (Figure 1). The CBM predicted clinical response with 73% sensitivity and 77% specificity. Cellworks CBM identified novel biomarkers responsible for platinum-based combination therapy response as mentioned below. +Etoposide: 13q-del, RB1-del, MBD1-del, LIG4-del, ERCC5-del, ATP7B-del +Gemcitabine: AKT3-amp, MAPKAP2-amp, TAP1-del +Vinorelbine: TET2-del/LOF, TRIB3-amp, SLX4-del +Paclitaxel: KLF4-del, SNCG-del, RAC1-amp/GOF +Docetaxel: BCL2L1-amp, HMGA1-amp, NSD1-del, SLC22A7-amp, FSIP1-del These genes contributed to drug efficacy by impacting various pathways, including DNA repair, oxidative-stress, methylation machinery, spindle formation, and mitotic-catastrophe. The aberration frequency of these genes was high among the responders within each subgroup and was very low in non-responders. Additionally, a model of clinical outcome versus the linear and quadratic function of efficacy score, drug combination and the interaction of both showed that efficacy score provides predictive information above and beyond the choice of drug combination alone (likelihood ratio chi-sq = 35.56, df=13, p-value = 0.0007). [Formula presented]
Conclusion(s): This pilot study highlights how the Cellworks CBM biosimulation platform can help identify patients for therapy response prediction. By using novel biomarkers, a CBM-informed decision tree can be employed to identify the optimal drug combination for platinum-based therapy. We suggest that this approach be validated prospectively in a larger patient cohort. Keywords: Cancer Therapy Biosimulation, Multi-omics Therapy Biosimulation, Personalized Cancer Therapy
Copyright

Introduction: KRAS is a frequent oncogenic driver in solid tumors, including Non-Small Cell Lung Cancer (NSCLC). KRAS is involved in various signaling pathways that could allow for targeting of KRAS by targeting downstream key transcription factors that mediate oncogene signaling. At the same time, co-occurrence of other mutations alters the signaling pathways and the key transcription factors involved in the disease network. The convergence of these dysregulated pathways to activate key kinases and transcription factors defines master regulators, forming the regulatory logic (i.e. oncotecture) of the tumor cell that maintains the malignant phenotype, its hallmark behaviors, and homeostasis in the face of treatment, while also providing a new set of potential therapeutic targets. In this study, we describe the impact of KRAS and a variety of co-mutations on the tumor's oncotecture.
Method(s): 248 NSCLC patients with KRAS mutations were selected from TCGA: 110 had only KRAS mutations, 138 had co-occurrence of other mutations: EGFR 55, MET 44, and PIK3CA 39. Mutation and CNV from each case served as input for the Cellworks Computational Omics Biology Model (CBM) to generate a patient-specific protein network map from PubMed and other resources. Disease-biomarkers unique to each patient were identified within protein network maps. The CBM identified the top 25 master regulators for each patient. We calculated the frequency of occurrence of each of the selected master regulators for KRAS alone and KRAS with other mutation co-occurrences.
Result(s): Comparative networks analyses of KRAS alone and KRAS with another mutation were performed. We identified that FOXM1, IKBKB, PLK1, PAK1, MTOR, AURKA, PIK3CA, CSNK2A1 function as master regulators in more than 70% of KRAS-only mutated cancers (Table 1). When co-mutations were present, the master regulator network was upregulated, e.g.: (1) MET: RPS6KA3, STAT3, NEK2 (2) EGFR: NFKB1, CEBPA, NEK2 (3) PIK3CA: SRPK1, GLI2, AKT [Formula presented]
Conclusion(s): The Cellworks CBM can reveal transcription factor addiction by identifying the convergence points of numerous upstream dysregulated pathways. These master regulators reveal another set of potential treatment vulnerabilities or Achilles heels in the network that can inform specific treatment options. This study identifies the key transcriptional mediators of KRAS mutations and how they are shuffled by the presence of co-mutations in other common oncogenes. The CBM biosimulation platform identifies the regulatory network in the cancer laying the foundation for new therapeutic strategies targeting key master regulators. Keywords: Personalized Cancer Therapy, Cancer Therapy Biosimulation, Multi-omics Therapy Biosimulation
Copyright

Introduction: PD-L1 is an immune checkpoint protein that mediates immune evasion. In Non-Small Cell Lung Cancer (NSCLC), its expression is used to predict the outcome of treatment targeting PD-1/L1. However, clinical benefits do not occur uniformly, and new approaches are needed to assist in selecting patients for immunotherapy.
Method(s): 26 patients with known clinical response to pembrolizumab were selected from publicly available data (PMID:25765070) (Table 1). Mutation and copy number aberrations from individual cases served as input into the Cellworks Omics Biology Model (CBM) to generate a patient-specific protein network map from PubMed and other online resources. Disease-biomarkers unique to each patient were identified within protein network maps. Digital drug biosimulations were conducted by measuring the effect of pembrolizumab on a cell growth score comprised of a composite of cell proliferation, viability, apoptosis, metastasis, and other cancer hallmarks. Drug biosimulations were conducted to identify and evaluate therapeutic efficacy.
Result(s): Among 26 patients treated with pembrolizumab, 14 were clinical responders, defined as stable disease or partial response lasting longer than 6 months, and 12 non-responders. Notably, 9/12 non-responders were PD-L1 positive (Table 1). Cellworks biosimulation predicted response with 84.6% accuracy, 75% specificity, and 92.86% sensitivity. Positive predictive value was 81.25% and negative predictive value was 90%. CBM identified that response was influenced by pathways that impacted the tumor microenvironment (TME). Deletions of adenosine pathway genes were observed in responders, whereas CNVs for these genes were enriched in non-responders (Table 1). Loss of ENPP1, and INSIG1 and SENP2 CNV aberrations, all of which regulate the STING pathway, could play a significant role in governing immune checkpoint blockade (ICB) response. Although STK11 loss appears to be a biomarker for poor ICB response, it was equally enriched in responders (n=7) and non-responders (n=7) in this dataset. Notably, responders had STK11 mutations and chromosome 6 loss, whereas non-responders had STK11 loss with chromosome 6 wild type or gain. Finally, frameshift mutations (FSM) enhance neoepitope formation and were higher in responders (average FSM = 48) vs non-responders (average FSM = 19). [Formula presented]
Conclusion(s): Alterations of the adenosine and STING pathways play key roles in determining benefit from PD-1/L1 targeting and highlight therapeutic possibilities for improving outcome in specific patient subgroups based on PD-L1 expression. The Cellworks CBM captures a holistic picture of the TME using tumor omics and improves response prediction beyond PD-L1 testing. Keywords: Multi-omics Therapy Biosimulation, Personalized Cancer Therapy, Immunotherapy Biosimulation
Copyright

Introduction: Sotorasib is a first-in-class small molecule that specifically and irreversibly inhibits KRAS^G12C. The phase 1/2 CodeBreaK 100 trial evaluated sotorasib in patients with pretreated advanced non-small cell lung cancer (NSCLC) harboring KRAS p.G12C. In the registrational phase 2 part, sotorasib showed an objective response rate (ORR) of 37.1% and a median progression-free survival (PFS) of 6.8 months. Here, we report on the activity of sotorasib in patients with treated brain metastases (BM).
Method(s): Patients from the phase 1/2 CodeBreaK 100 trial receiving 960mg dose were included. Patients with active untreated BM were excluded. Patients who had BM resected or had received radiation therapy ending >=4 weeks prior to the trial were eligible. Systemic response was assessed by independent central review per RECIST 1.1. The presence of neurologically stable/asymptomatic BM at baseline was determined by investigators. CNS response was retrospectively evaluated by central neuroradiologic review, using the response assessment in neuro-oncology BM (RANO-BM) criteria, in patients with >=1 target CNS lesions (>= 10mm) and/or non-target CNS lesions. For non-target lesions, stable disease (SD) refers to response that is neither complete response (CR) nor progressive disease (PD).
Result(s): 174 patients were included: 40 had stable BM (23.0%) while 134 (77.0%) had no BM at baseline. In the BM group, 65% had received prior radiotherapy, and 20% had received prior brain surgery. Systemic efficacy of sotorasib per RECIST 1.1 is shown in the Table. Per central RANO-BM review, 16 patients had baseline and >=1 on-treatment evaluable scans: 3 had target and 13 had non-target CNS lesions. 9 patients had 1 lesion, 2 had 4 lesions, and 5 had >=5 lesions. Of 13 patients with non-target CNS lesions, 2 had CR, 11 had SD. Of 3 patients with target lesions, 1 had SD, and 2 had PD. Overall, intracranial disease control was achieved in 14 of 16 patients (87.5%) with evaluable BM. Safety in the BM group was consistent with previous reports. [Formula presented]
Conclusion(s): Sotorasib demonstrated systemic durable anticancer activity, with a median PFS and OS of 5.3 and 8.3 months in NSCLC patients with stable BM previously treated with radiation or surgery. Intracranial complete responses were observed, with continued intracranial stabilization observed in the majority of patients with evaluable BM. Additional studies are ongoing to evaluate sotorasib in patients with active untreated BM (NCT04185883). Keywords: brain metastases, KRAS p.G12C, sotorasib (AMG 510)
Copyright

Background: Time on treatment, also called time to treatmentdiscontinuation, is a readily available real-world effectiveness endpointhighly correlated at the patient-level with progression-free survival andmoderately to highly correlated with overall survival in clinical trials andreal-world data. In October 2016, pembrolizumab received FDA approvalbased on results from KEYNOTE-024, as a first-line monotherapy forpatients with metastatic NSCLCwith PD-L1 tumor proportion score (TPS)>=50%andnoEGFR/ALKgenomicaberrations, administereduntil diseaseprogression, unacceptable toxicity, or up to 24 months. In KEYNOTE-024,25%(39/154) of patients received 35 cycles (2 years) of pembrolizumabas initially assigned therapy. Our objective was to describe rwToT withfirst-line pembrolizumab in real-world oncology practice.
Method(s): Using the US nationwide Flatiron Health electronic healthrecord-derived, de-identified database, we included adult patients withpathologically confirmed advanced, PD-L1 TPS >=50% NSCLC whoinitiated first-line pembrolizumab monotherapy from November 2016-September 2019, with follow-up through September 2020. Eligibilitycriteria included ECOG performance status 0-2, PD-L1 TPS >=50%, noEGFR/ALK genomic aberration, and no known ROS1 aberration. Patientsenrolled in a clinical trial were excluded. Median rwToT and landmarkon-treatment rates were estimated using Kaplan-Meier method.
Result(s):(Table Presented)
Conclusion(s): Patients with key trial-eligible characteristics (ECOG 0-1,PD-L1 TPS >=50%, EGFR/ALK negative) experienced rwToT with firstlinepembrolizumab similar to the phase III pivotal clinical trial.Approximately 23% received at least 2 years of treatment, suggestinglong-term benefit of pembrolizumab monotherapy for PD-L1 TPS >=50%advanced NSCLC in a real-world setting.
Copyright International Association for the Study of Lung Cancer. Published by Elsevier Inc

Background Immune checkpoint inhibitors (ICI) have revolutionized the management of lung tumors decreasing mortality rates. However, the response rates to these ICI drugs are limited, and identifying those patients who are most likely to benefit remains a clinical challenge. Due to the complex nature of the host immune response, tissue-based biomarker development for immunotherapy (IO) is challenging. Consequently, there is a critical unmet need to develop accurate, validated imaging biomarkers to predict which Non-Small Cell Lung Cancer (NSCLC) patients will benefit from IO. Airway deformations such as central airway obstruction can be considered an important manifestation of cancer aggressiveness or metastatic disease and may have a significant impact on therapeutic refractoriness. In this study, we sought to evaluate whether quantitative measurements of lung airway morphology (QuaLM) on baseline CT scans are associated with response and overall survival in NSCLC patients treated with ICI. Methods In this retrospective study, 80 cases who underwent 2-3 cycles of PD1/PD-L1 ICI therapy (nivolumab/pembrolizumab/ atezolizumab) were included. RECIST v1.1 was used to define 'responders' and 'non-responders'. Patients were randomly divided into a training (n=40) and a test set (n=40). A region growing algorithm is applied to the trachea, identified by Hough transform, to segment bronchi and bronchioles (figure 1a). 14 QuaLM features were extracted from segmented airway on CT scans. Wilcoxson ranksum test is used to identify the predictive QuaLM features. The top 4 QuaLM features in conjunction with a linear discriminant machine learning classifier were used to predict the response to IO. We also built a QuaLM risk score using the least absolute shrinkage and selection operator (LASSO) Cox regression model to predict overall survival (OS). Results The response prediction model trained with top QuaLM features (table 1) predicts responders to ICI with an area under research operating characteristic curve (ROC AUC) of 0.67+/-0.08 (figure 1.b) in the training (St) and AUC=0.63 in the test set (Sv). The airway radiomics risk-score was found to be significantly associated with OS in St (HR=2.34, 95% CI:[1.08-5.07], P=0.008) and Sv (HR=2.55, 95% CI:[0.8- 8.1], P=0.034) (figure 1.c). Conclusions QuaLM features were able to distinguish responders from non-responders and also were found to be associated with OS for NSCLC patients treated with ICI. With additional validation, QuaLM could potentially serve as an imaging biomarker of ICI response assessment for NSCLC patients. This could allow the selection of NSCLC patients who will benefit from IO and help design more rational clinical trials with a combination of IO

OBJECTIVE:Patients with non-small cell lung cancer (NSCLC) metastatic to the brain are living longer. The risk of new brain metastases when these patients stop systemic therapy is unknown. The authors hypothesized that the risk of new brain metastases remains constant for as long as patients are off systemic therapy. METHODS:A prospectively collected registry of patients undergoing radiosurgery for brain metastases was analyzed. Of 606 patients with NSCLC, 63 met the inclusion criteria of discontinuing systemic therapy for at least 90 days and undergoing active surveillance. The risk factors for the development of new tumors were determined using Cox proportional hazards and recurrent events models. RESULTS:The median duration to new brain metastases off systemic therapy was 16.0 months. The probability of developing an additional new tumor at 6, 12, and 18 months was 26%, 40%, and 53%, respectively. There were no additional new tumors 22 months after stopping therapy. Patients who discontinued therapy due to intolerance or progression of the disease and those with mutations in RAS or receptor tyrosine kinase (RTK) pathways (e.g., KRAS, EGFR) were more likely to develop new tumors (hazard ratio [HR] 2.25, 95% confidence interval [CI] 1.33-3.81, p = 2.5 × 10-3; HR 2.51, 95% CI 1.45-4.34, p = 9.8 × 10-4, respectively). CONCLUSIONS:The rate of new brain metastases from NSCLC in patients off systemic therapy decreases over time and is uncommon 2 years after cessation of cancer therapy. Patients who stop therapy due to toxicity or who have RAS or RTK pathway mutations have a higher rate of new metastases and should be followed more closely.